=Paper=
{{Paper
|id=Vol-2104/paper_221
|storemode=property
|title=Deterministic and Stochastic Models of Decision Making in Air Navigation
Socio-Technical System
|pdfUrl=https://ceur-ws.org/Vol-2104/paper_221.pdf
|volume=Vol-2104
|authors=Tetiana Shmelova,Yuliya Sikirda,Claudio Scarponi,Antonio Chialastri
|dblpUrl=https://dblp.org/rec/conf/icteri/ShmelovaSSC18
}}
==Deterministic and Stochastic Models of Decision Making in Air Navigation
Socio-Technical System==
https://ceur-ws.org/Vol-2104/paper_221.pdf
Deterministic and Stochastic Models of Decision Making
in Air Navigation Socio-Technical System
Tetiana Shmelova1[0000-0002-9737-6906], Yuliya Sikirda2[0000-0002-7303-0441], Claudio
Scarponi3[0000-0001-9022-7038], Antonio Chialastri4[0000-0001-5692-7161]
1
National Aviation University, Komarova av., 1, 03058, Kiev, Ukraine
shmelova@ukr.net
2
Flight Academy of the National Aviation University,
Dobrovolskogo Street, 1, 25005, Kropivnitsky, Ukraine
sikirdayuliya@ukr.net
3
Sapienza University of Rome, Italy
claudio.scarponi@uniroma1.it
4
Centro Studi Trasporto Aereo Sicurezza & Ambiente (CSTASA), Italy
antonio.chialastri@gmail.com
Abstract. The conceptual model of System for control and forecasting the
emergency situations development that taking into account the influence on de-
cision making by Air Navigation Socio-Technical System human-operator of
professional and non-professional factors has obtained. Optimization models of
decision making such as deterministic model, stochastic model (under risk and
uncertainty), neural network models have presented. Behavioral models appro-
priate using in Decision Support Systems for timely predicting of human-
operator actions in the emergencies.
Keywords: decision making, deterministic and stochastic models, human-
operator, neural network, professional and non-professional factors.
1 Introduction
At present, one of the main strategic problems of humanity on the path to sustaina-
ble development is the safety and reliability of technogenous production, which is a
complex system of interconnected technical, economic and social objects; has a multi-
level hierarchical structure and characterized by a high risk [1]. Emergencies, catas-
trophes, accidents in hydraulic engineering, chemical and military industries, gas and
oil pipelines, nuclear power plants, as well as in transport are frequent and common-
place Air Navigation System (ANS), in which there is a close interaction between
man and technological components, also evolved towards integrated Socio-Technical
Systems (STS) [1, 2]. Socio-technical systems, as a rule, have two common features:
the presence of hazardous activities and the use of high technology.
It is believed that aviation is the safest type of mass transportation and one of the
safest socio-technical production systems in the history of humanity. For a century,
aviation has gone a long way in the field of safety of flights from an unstable system
to the first "ultra-safe" system in the history of transport, that is, a system in which the
number of catastrophic failures in the field of safety is less than one million produc-
tion cycles [3].
Over the year 2017, the Aviation Safety Network [4] has recorded a total of 10 fa-
tal aircraft accidents, resulting in 44 occupant fatalities and 35 persons on the ground.
This makes 2017 the safest year ever, both by the number of fatal accidents as well as
in terms of fatalities. In 2016 Aviation Safety Network has recorded 16 accidents and
303 lives lost. Five accidents involved cargo flights, five were passenger flights. Giv-
en the expected worldwide air traffic of about 36,8 million flights, the accident rate is
one fatal passenger flight accident per 7,36 million flights. Since 1997, the average
number of aircraft accidents has shown a steady and persistent decline due to the con-
tinuing flight safety-driven efforts by international aviation organisations.
Elimination of accidents remains the key point for all kinds of aviation activity.
But it is impossible for aviation systems to be completely free of hazardous factors
and associated with them risks. Neither human activity nor human designed systems
are completely free of operational errors and its consequences [3, 5]. Flight safety is a
dynamical parameter of aviation system. Thus, risk factors should continuously miti-
gate. It is important to note that the adoption of indicators for the effectiveness of
safety of flights is often influenced by internal and international standards, as well as
cultural features [6]. When the risk factors and operational errors are reasonably mon-
itored, flight safety can be managed [3].
Statistics over the past decades indicate the dominant role of the human factor in
the total number of aviation accidents, which is about 80% [5, 7]. Therefore, re-
searches of the human factor effect on flight safety remain relevant.
The circular of ICAO presents the safety cases for cultural interfaces in aviation
safety with reference to established main conceptual safety models: SHEL model,
Reason’s model of latent conditions, Threat and Error Management (TEM) model and
other human factor's models [1, 3]. There are four stages of the evolution of the hu-
man factor's models (from 1972 to present time) [8, 9] associated with the appearance
of new system components and the diagnosis of human-operator (H-O) errors:
1. Professional skills of H-O / Interaction of H-Os / Definitional of H-O’s errors.
2. Cooperation in team / Interaction of H-Os in team / Error detection.
3. Influence of Culture / Safety / Error prevention.
4. Safety Management / Safety balance models / Minimization of errors.
The component Culture means the ongoing interaction of a group of people with
their environment. Culture develops and changes due to technological, physical, and
social changes in the environment. When pursuing safety in these systems, it is nar-
row and restrictive to look for explanations for accidents or safety deficiencies exclu-
sively in technical terms or purely from the perspective of the behavioral sciences. It
is necessary to systematically analyze and classify all the factors in the ANS as STS
that have an impact on the H-O in the performance of professional activities guided
by the requirements of ICAO documents in accordance with the steps below:
1. Analysis of ANS as STS: diagnostics, monitoring of the all factors (professional
and non-professional (individual-psychological, socio-psychological and psychophys-
iological factors) that influence on decision making (DM) by the H-O in STS.
2. Determining the professional type of the operators namely energy consumption
for the choice of profession and the compatibility of operators in the group.
3. Complex accounting of the all factors affecting the operator's DM in the STS.
4. Modelling of DM in STS using deterministic and stochastic models of DM in
STS by an H-O in emergence situations (ES) (under conditions of stochastic reflexive
bipolar choice too); neural network, Markov, GERT (Graphical Evaluation and Re-
view Technique)-models of DM in STS; models of diagnosis of the emotional state of
H-O in ES, etc.
5. Forecasting the emergency. Preventing the catastrophic situations.
Therefore, taking into account the influence on DM process by ANS H-O the pro-
fessional factors (knowledge, habits, skills, experience) as well as the factors of non-
professional nature (individual-psychological, psycho-physiological and socio-
psychological) [10–12] will allow to predict the H-O’s actions on the basis of the
reflexive theory [13].
The purposes of the work are: formalization of the influence of the professional
and non-professional factors on the H-O DM within ANS as STS; development of
models of H-O DM in Air Navigation Socio-Technical System (ANSTS); working-
out computer programs for Decision Support Systems (DSS) of H-O in ES.
2 Optimization Models of Decision Making by Human-
Operator in Air Navigation Socio-Technical System
Decomposition of the DM process by H-O ANS and the systemic analysis of influ-
ence of the factors of professional and non-professional activities on the DM in
ANSTS were done [10, 11]. In order to take into account the complex of the factors
that influencing on H-O of the ANSTS in the expected and unexpected conditions of
an aircraft operation a reflexive model of bipolar choice of H-O was worked-out [8].
The conceptual model of System for control and forecasting the ES development
that using DM models on the base of Artificial Intelligence System (AIS) / Decision
Support System (DSS) was obtained (Fig. 1), where F p F ed , F exp – are the pro-
fessional factors; F np F ip , F pf , F sp – are the non-professional factors; F ed – are
the knowledge, skills and abilities, acquired H-O during training; F exp – are the
knowledge, skills and abilities, acquired H-O during professional activity;
F ip f ipt , f ipa , f ipp , f ipth , f ipi , f ipn , f ipw , f iph , f exp – is a set of H-O individual-
psychological factors (temperament, attention, perception, thinking, imagination,
nature, intention, health, experience); F pf – is a set of H-O psycho-physiological
factors (features of the nervous system, emotional types, sociotypes);
F sp f spm , f spe , f sps , f spp , f spl – is a set of H-O socio-psychological factors (mor-
al, economic, social, political, legal factors).
The analysis of social-physiological factors conducted by the authors allowed to
make a conclusion that the activities of pilots are influenced by the own image, the
image of corporation as well as by interests of a family. At the same time respondents
– air traffic controllers (ATC) pay special attention to interests of their families, their
own economic status and professional promotion [11, 12].
Fig. 1. System for control and forecasting the ES development
Deterministic and stochastic models for ANS H-O (pilot, controller) were obtained
in accordance with the flight manual of aircraft or the adopted technologies of con-
troller’s work ASSIST (Acknowledge, Separate, Silence, Inform, Support, Time) in
ES. Deterministic and stochastic models for ATC are presented in Fig. 2, where {А} –
is the set of the operations which are carried out by the controller in accordance with
ASSIST; {Т} – is the time of decision making; {Р} – is the set of the probabilities of
j-factor influence during i-alternative solution choice; {U} – is the set of the losses
associated with choosing i-alternative solution during j-factor influence; {R} – is the
set of the risks associated with choosing i-alternative solution during j-factor influ-
ence; {λ} – is the set of the factors influencing DM.
a) b)
Fig. 2. Models of DM in ANSTS: а) deterministic model; b) stochastic model
With using neural network models, the values of probabilities (pn) [11], expected
outcomes (rk) and additional inputs - factors (ξk) (Fig. 3) of ES development were
received.
Fig. 3. Neural network model of ES development with additional inputs of influencing DM
factors.
The network has additional inputs, called the Bias (offset) that takes into account
additional restrictions on calculating parameters (2):
n
p u 0.
i 1
i i k (2)
where рi – are the weight coefficients; uі – are the neural network inputs; ξk– is a
Bias (shift) under influencing factors of uncertainty (Table 1).
Table 1. Matrix of Bias identification
Factors that influencing on the decision making
Alternative decisions
λ1 λ2 ··· λj ··· λm
А1 ξ11 ξ12 ··· ξ1j ··· ξ1m
А2 ξ21 ξ22 ··· ξ2j ··· ξ2m
··· ··· ··· ··· ··· ··· ···
Аі ξi1 ξi2 ··· ξij ··· ξim
··· ··· ··· ··· ··· ··· ···
Аn ξn1 ξn2 ··· ξnj ··· ξnm
The outcomes of neural network are (3):
R f ( net ) , (3)
where f – is a non-linear function (active function) that takes into account the
time of decision making tі; net – is a weighted sum of inputs.
The optimal solution is found by the criterion of an expected value with the Sav-
age criterion (4):
n
Aopt min maxR min max ti (
i 1
pi ui k ).
(4)
The critical time of the flight crew actions in case of an engine failure on take-off
and approach to land in the bad weather conditions was obtained [10]. The selection
in the direction of the negative pole leads to the maximum expected risk R=1028. The
choice in the direction of the positive pole when the ES occurs at the first stage of DM
by H-O ANS (for example, a flight to alternative aerodrome) has a risk which is 60,5
times lesser: R=17.
In stochastic network of the flight situation development of GERT type the tops are
represented by stages of the situation (normal, complicated, difficult, emergency or
catastrophic), and the arcs are represented by a process of transition between stages of
the situation. The algorithm of stochastic network analysis was developed [10, 11].
Thus according to results of stochastic network analysis of the flight situation devel-
opment from normal to catastrophic the following values obtained: mathematical
expectation of flight situation development time tij – М[tij]; the variance of flight sit-
uation development time tij – [tij]; the probability of flight situation development pij
2
– рij,.рji, рiі. Based on the W-functions of positive and negative of H-O choice the Mar-
kov's network of flight situations' development from normal to catastrophic was con-
structed [8].
In addition, with using reflexive model the risks RA, RB of DM in the ANS under
the influence of the external environment x1, the previous H-O’s experience x2 and the
intentional choice of H-O x3 have obtained [8]. The expected risk in the process of
DM of H-O is equal (5):
R A min Rij
, (5)
RDM RB ,
R
AB X ( x1 , x2 , x3 ), ,
where RА – is an expected risk of the DM for H-O with taking into account the cri-
terion of the expected value minimization; RВ – is an expected risk of the DM for H-O
with taking into account his model of preferences; Rij – is an expected risk for making
Аij-decision; γ – is a concept of a rational individual’s behaviour; ρ – is a system of
individual’s preferences in a concrete situation of the choice; RАВ – is a mixed choice
made by a H-O.
For example, if the pilot, the ATC and the society have a choice in the direction of
the negative pole B, the preferences model can form the plane of the disaster K [8].
Methodology of research and training in ANS as STS has developed [14]. Let’s
consider the individual works of aviation students and post-graduate students in edu-
cation (course “Basic of DM in ANS” in National Aviation University, Kyiv) after
Master class of DM in ANS [15].
Research has shown that the choice of the optimal variant of the forced flight com-
pletion in emergencies requires from the operator to analyze the significant amount of
diverse information. The following conceptual models of DSS in ANS have obtained,
such as DSS for ATC in emergencies, for example “Aircraft Decompression”, “Low
oil pressure”, “Engine failure”, etc. [10, 11, 15]; DSS for flight dispatcher for support
of the DM regarding aircraft landing in emergencies to choice alternative landing
aerodrome [10, 11]; DSS for operator of Unmanned Aerial Vehicles (UAV) in emer-
gencies situation, for example in losing of communication with UAV and choosing
optimal landing place, etc. [16]. DSS contain common sets of components, such as
data related components, algorithm related components, user interface and display
related components. The user interface and the result of calculation of DM process by
H-O (pilots, ATC, UAV’s operators) under risk are presented in Fig. 4 [17]. With
using this program operator can obtain optimal solution for such problem as landing
in bad weather condition, ES in flights, etc.
Fig. 4. The result of calculation of H-O DM process under risk.
3 Conclusion
The conceptual model of System for control and forecasting the ES development that
taking into account the influence on DM process by ANS H-O of the professional
factors (knowledge, habits, skills, experience) as well as the factors of non-
professional nature (individual-psychological, psycho-physiological and socio-
psychological) has presented. Deterministic and stochastic models for ANS H-O (pi-
lot, controller) have obtained in accordance with the flight manual of aircraft or the
adopted technologies of ATC work ASSIST. With using neural network model, the
values of probabilities of ES development have received. The optimal solution has
found by the criterion of an expected risk minimization.
Further research should be directed to solution of the complex practical tasks of
improving the operator’s actions in different cases of emergencies, to creation the
software for these problems. Models of ES development and of DM by UAV’s in ES
will allow predicting the H-O’s actions with the aid of the informational-analytic and
diagnostics complex for research H-O behavior in extreme situation.
References
1. Human Factors Guidelines for Safety Audits Manual: Doc. ІСАО 9806-AN/763 (1st ed.).
International Civil Aviation Organization, Canada, Montreal (2002)
2. Campbell, R.D., Bagshaw, M.: Human Performance and Limitations in Aviation (3-rd ed.).
Blackwell Science Ltd, UK (2002)
3. Safety Management Manual (SMM): Doc. ICAO 9859-AN/474 (3rd ed.). International
Civil Aviation Organization, Canada, Montreal (2013)
4. Aviation Safety Network: ASN data show 2017 was safest year in aviation history,
https://news.aviation-safety.net/2017/12/30/preliminary-asn-data-show-2017-safest-year-
aviation-history, last accessed 03.03.2018
5. Kylik, N.S., Kharchenko, V.P., Lytsky, M.G., etc.: Encyclopedia of Аviation Safety.
Technology, Kyiv (2008)
6. Cross-Cultural Factors in Aviation Safety: Human Factors Digest Nо. 16: Сirc. ІСАО 302-
AN/175. International Civil Aviation Organization, Canada, Montreal (2004)
7. Leychenko, S., Malishevskiy, A., Mikhalic, N.: Human Factors in Aviation: monograph in
two books. Book 1st. YMEKS, Kirovograd (2006)
8. Shmelova, T., Sikirda, Y., Rizun, N., Salem, A.-B. M., Kovalyov, Y.: Socio-Technical De-
cision Support in Air Navigation Systems: Emerging Research and Opportunities:
monusript. International Publisher of Progressive Information Science and Technology Re-
search, USA, Pennsylvania (2017)
9. Shmelova, T., Scarpony, C., Chialastri, A.: Evolution of Human Factor's Models. Іn:
Ukrainian Scientific and Practical Conference "Problems of Navigation and Control of the
Movement", рр. 56. NAU, Kyiv (2017)
10. Kharchenko, V., Shmelova, T., Sikirda, Y.: Decision-Making in Socio-Technical Systems:
monograph. NAU, Kyiv (2016)
11. Kharchenko, V., Shmelova, T., Sikirda, Y.: Decision-making of operator in Air Navigation
System: monograph. KFA of NAU, Kirovograd (2012)
12. Rizun, N., Shmelova, Т.: Decision-Making Models of the Human-Operator as an Element
of the Socio-Technical Systems. Іn: Strategic Imperatives and Core Competencies in the
Era of Robotics and Artificial Intelligence, рр. 167–204. International Publisher of Pro-
gressive Information Science and Technology Research, USA, Pennsylvania (2016)
13. Lefebvre, В., Adams-Webber, J.: Functions of Fast Reflexion in Bipolar Choice. Reflexion
Processes and Control, vol. 1, рр. 29–40 (2002)
14. Kharchenko, V., Shmelova, T., Sikirda, Y.: Methodology of Research and Training in Air
Navigation Socio-technical System. Proceedings of the National Aviation University, vol.
1, рр. 8–23 (2018)
15. Shmelova, T., Mirskova, M., Krepak D.: Decision Making in Еmergency Situation of Air
Navigation System’s Оperator. Іn: ХІІІ International Siencе-Practical Conference "AVIA-
2017", рр. 11.53-11.59. NAU, Kyiv (2017) –
16. Kharchenko, V., Shmelova, T., Bondarev, D., Stratiy, A.: Modelling of Decision Making
of Unmanned Аerial Vehicle's Оperator in Еmergency Situations. Proceedings of the Na-
tional Aviation University, vol. 1, рр. 20–28 (2017)
17. Shmelova, T., Znakovska, E.: Computer program "Decision Tree": author's certificate.
State Service of Intellectual Property of Ukraine, Kyiv (2015)